Hacker News Reader

I had access to GitHub Copilot as a student in early 2022 while learning Haskell and immediately realised that it would hinder my learning if I didn't turn it off and implicitly follow this understand, plan, execute, reflect loop.

AI products like Cursor have the notion of an 'autonomy slider' [1] that can fortunately be turned all the way down (disable Cursor Tab) but relying on this discipline seems fickle when with the right agentic loops [2] and context engineering, thousands of lines of code can be churned out with minimal supervision.

I've considered always working on two projects over a long timespan, one with no AI assistance, possibly in a separate IDE like Zed, and one in Vibe Kanban (my current daily driver) but this feels like an inefficient proxy to accelerating this four step learning loop with a tool like solveit.

Since the solveit product isn't released and seemingly isn't competing with solutions, is there an opportunity to convey how AI product developers should be thinking about amplifying their users and keeping them in the learning loop?

So far, I've seen Claude Code's Learning output style [3], and also ChatGPT's study mode but in these cases, the only product change is a prompt and solveit is more than that.

[1] https://www.latent.space/i/166191505/part-a-autonomy-sliders [2] https://simonwillison.net/2025/Sep/30/designing-agentic-loop... [3] https://docs.claude.com/en/docs/claude-code/output-styles#bu...

britannioon MCP: An (Accidentally) Universal Plugin System

I discovered the "Layered Tool Pattern" a few days ago which exposes three tools to discover and execute an arbitrary number of service endpoints. https://engineering.block.xyz/blog/build-mcp-tools-like-ogre...

britannioon Spaced Repetition Memory System

I used this throughout my computer science degree, and it worked a charm. Now I rely on it as a solid personal knowledgebase.

britannioon 12-factor Agents: Patterns of reliable LLM applications

looking forward to the new tool

britannioon Rama, the 100x developer platform, is now free for production use

> We develop your entire backend according to your spec for 75% less cost than anyone else charges for a comparably high-quality implementation

Icing on the cake

britannioon Fitness Trackers Are Only 67% Accurate, New Research Finds

Why didn't they include Whoop or Oura?

britannioon Snyk security researcher deploys malicious NPM packages targeting cursor.com

The lack of a bug bounty program doesn't prohibit them from rewarding reported vulnerabilities.

britannioon Making AMD GPUs competitive for LLM inference (2023)

Some of the work from Tinycorp is: https://github.com/tinygrad/7900xtx

britannioon How we made our AI code review bot stop leaving nitpicky comments

Only while a vendor is ahead of the others. We developers will favour a vendor with faster inference and lower pricing.

britannioon Tldraw Computer

It was a cool thing... I expected a hacky demo that'd fall apart mid-way but it held up. The Macintosh SE in the office was cool too.

britannioon Clay – UI Layout Library

So cool. I wonder if it'd work for a Raspberry Pi Pico + https://pimoroni.com/picodisplay or similar devices.

britannioon We're forking Flutter

The data is supposedly there: https://news.ycombinator.com/item?id=41975320

britannioon Collaborative text editing with Eg-Walker: Better, faster, smaller

Joseph explains the algorithm on YouTube too: https://www.youtube.com/watch?v=rjbEG7COj7o

It's great work, combining the best of OT and CRDTs.

britannioon Rustpad is an efficient and minimal open-source collaborative text editor

You can combine the best of both too: https://www.youtube.com/watch?v=rjbEG7COj7o

britannioon Launch HN: Silurian (YC S24) – Simulate the Earth

Andrej Karpathy states that LLMs are highly general purpose technology for statistical modelling of token streams [1]. For example, comma.ai uses transformers in their self-driving model which is far from a linguistic task.

[1] https://x.com/karpathy/status/1835024197506187617 [2] https://www.youtube.com/watch?v=-KMdo9AWJaQ&t=1010s

britannioon Faster CRDTs (2021)

But in the eg-walker paper, you benchmark against Yjs rather than Yrs because Yrs performed worse?

britannioon Friend.com – The Virtual Friend

It's actually the complete opposite: https://x.com/AviSchiffmann/status/1818460679680278709

britannioon Show HN: 40M embeddings to find who knows what on HN

Text selection doesn't work, and I can't navigate to the user's news.ycombinator.com profile; otherwise, it's neat.

britannioon Crafting Interpreters

I just finished the second half, it's a great book! I recommend doing one or two of the proposed challenges in each chapter to reinforce your understanding of the content.

britannioon Local First, Forever

You can charge the user once upfront, possibly after a free trial. 37 signals are trying it out with once.com. As a developer, you are in control of how local-first your application should be.

britannioon Local First, Forever

Matthew Weidner's blog post has been enlightening [1]. When a centralised server is involved, CRDTs and OT are merely optimisations over server reconciliation so yes CRDTs aren't always the best approach. It is a compelling approach for text editing though.

[1] https://mattweidner.com/2024/06/04/server-architectures.html